Getting Started in R and Rstudio

Parts 1 & 2

Author

Affiliation

Jessica Minnier, Meike Niederhausen

OHSU OCTRI-BERD

Published

September 19, 2024

1 Load files for today’s workshop

Open Google Doc https://bit.ly/berd_r_intro_2024_doc
- Use this to post questions!
- This has all the links in it
Download zipped folder of files using link https://bit.ly/berd_intro_2024_files
- Do you have a PC? Need to right click .zip file and select EXTRACT ALL
- Do you have a Mac? Just double click .zip file and it will open
- These files are hosted on github: https://github.com/OHSU-OCTRI-BERD/R_Intro_2024_09
Slides are here: https://bit.ly/berd-r-intro2024-slides or you can open the slides file (html) in a web browser: slides_R_berd_2024.html
Open RStudio by double-clicking on the file berd_r_intro_2024_09.Rproj

1.1 Overview

1.1.1 Part 1: 09/19/2024

Intro to R & RStudio
Quarto documents & code chunks
R packages
Data: loading, viewing
Summarizing data
- Quantitative
- Categorical
Data visualizations
- Quantitative: histograms, boxplots, scatterplots
- Categorical: barplots
Missing data summary

1.1.2 Part 2: 10/03/2024

Data wrangling
- Sorting data
- Subsetting data (rows, columns)
- Creating new variables in a dataset
- Factor level variables
Some more data viz (stratifying figures)
Replacing missing values with another value
Resources

2 R Packages: install & load

2.1 Install packages used in this workshop

Run the code below to install packages:

install.packages("tidyverse")  # this takes a while to install
install.packages("writexl")
install.packages("rstatix") 
install.packages("janitor")
install.packages("naniar")
install.packages("gt")
install.packages("gtsummary")

It is VERY IMPORTANT that you install these packages in order to run the code for the workshop!

2.2 Load packages with `library()` command

Use the library() command to load each required package.
Packages need to be loaded every time you open Rstudio.
- The code to load packages MUST be in your qmd file.
- Tip: at the top of your qmd file, create a chunk that loads all of the R packages you want to use in that file.

library(janitor)    # run this every time you open Rstudio


Attaching package: 'janitor'

The following objects are masked from 'package:stats':

    chisq.test, fisher.test

Check to see whether a package has been loaded:
- Go to Packages tab
- If box to left of package name is checked off, then it’s loaded

2.3 Load packages used in this workshop

Run the code below in your R session

library(tidyverse)

── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
✔ dplyr     1.1.4     ✔ readr     2.1.5
✔ forcats   1.0.0     ✔ stringr   1.5.1
✔ ggplot2   3.5.1     ✔ tibble    3.2.1
✔ lubridate 1.9.3     ✔ tidyr     1.3.1
✔ purrr     1.0.2     
── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag()    masks stats::lag()
ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors

library(writexl)
library(rstatix)


Attaching package: 'rstatix'

The following object is masked from 'package:janitor':

    make_clean_names

The following object is masked from 'package:stats':

    filter

library(janitor)
library(naniar)
library(gt)
library(gtsummary)
library(readxl)  # part of tidyverse, but not the tidyverse core

Note that the package names do not need quotes around them when using the library() function, but they are required when using the install.packages() function.

3 Data basics in R

3.1 Dataset example (toy data)

Use Excel to open the dataset toy_data.xlsx that is located in the data folder.

3.2 Load data from an .xlsx (Excel) file: point & click option

In RStudio’s Files window, open the data folder and
- then click on the dataset toy_data.xlsx.
Select the Import Dataset... option,
- Import Options: Type in NA in the “NA:” box, and
- then click on the Import button on the bottom right of the pop-up window.
Copy the code from the console used to load the file into your script (omitting the + sign):

toy_data <- read_excel("data/toy_data.xlsx", na = "NA")

3.3 Load data from an .xlsx (Excel) file: code option

The point and click option used the read_excel command from the readxl package to load the xlsx file.
When loading the dataset, we need to specify where the xlsx file is located, which in this case is in the folder called data.
No matter which option we use, the code to load the data MUST be in your qmd.

# library(readxl) # commented out since loaded above already
toy_data <- read_excel("data/toy_data.xlsx", na = "NA")

This code assigns the output of the read_excel() function (data) into an object named toy_data
The <- is the assignment operator
After loading, the name of the data frame object toy_data will be in your Environment tab

3.4 View the `toy_data` dataset

Run the View command below within RStudio.

View(toy_data)

A new tab in the scripting window will appear with the toy_data dataset.
Note that the code chunk in the qmd file specifies #| eval: FALSE so that this code is not run while rendering the qmd file.

3.5 What are the variable “types”?

R variable type	Description
`dbl`: double	numbers; also `num` or `int` for numbers or integers
`chr`: character	text, “strings”
`fct`: factor	categorical variables stored with levels (groups)
`lgl`: logical	boolean (TRUE, FALSE)

glimpse(toy_data)

Rows: 20
Columns: 11
$ id                     <dbl> 335340, 638618, 922382, 923122, 923963, 925603,…
$ age                    <chr> "17 years old", "16 years old", "14 years old",…
$ sex                    <chr> "Female", "Female", "Male", "Male", "Male", "Ma…
$ grade                  <chr> "10th", "9th", "9th", "9th", "10th", "10th", "1…
$ race4                  <chr> "White", NA, "White", "White", "Black or Africa…
$ bmi                    <dbl> 27.5671, 29.3495, 18.1827, 21.3754, 19.5988, 22…
$ weight_kg              <dbl> 66.23, 84.82, 57.61, 60.33, 63.50, 70.31, 45.36…
$ text_while_driving_30d <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, "0 days…
$ smoked_ever            <chr> NA, "Yes", "Yes", "Yes", "No", "No", "Yes", "No…
$ bullied_past_12mo      <lgl> NA, NA, FALSE, FALSE, TRUE, TRUE, TRUE, FALSE, …
$ height_m               <dbl> 1.550000, 1.699999, 1.779999, 1.680001, 1.79999…

3.6 Getting to know the dataset

3.6.1 How many rows and columns?

# dim is for dimensions
# Output: # of rows, # of columns
dim(toy_data)

[1] 20 11

nrow(toy_data)

[1] 20

ncol(toy_data)

[1] 11

3.6.2 What are the variable names?

names(toy_data)

 [1] "id"                     "age"                    "sex"                   
 [4] "grade"                  "race4"                  "bmi"                   
 [7] "weight_kg"              "text_while_driving_30d" "smoked_ever"           
[10] "bullied_past_12mo"      "height_m"

4 The Youth Risk Behavior Surveillance System (YRBSS)

Data from the CDC’s Youth Risk Behavior Surveillance System (YRBSS)

complex survey data
national school-based survey
- conducted by CDC and state, territorial, and local education and health agencies and tribal governments
monitors six categories of health-related behaviors
- that contribute to the leading causes of death and disability among youth and adults
- including alcohol & drug use, unhealthy & dangerous behaviors, sexuality, and physical activity
- see Questionnaires
the data in yrbss_demo.xlsx are a subset of data in the R package yrbss, which includes YRBSS from 1991-2013

5 Practice time! (part 1)

Use R code to answer the following questions.

Open the Excel file yrbss_demo.xlsx in the data folder to familiarize yourself with it.
Load the dataset yrbss_demo.xlsx. Make sure the name of the loaded data is yrbss_data.

What are the column names?

How many rows and columns are in yrbss_data?

Which variables are numeric (quantitative)?

Which variables are character?

6 Load the YRBSS data

Open the Excel file yrbss_demo.xlsx in the data folder to familiarize yourself with it.
This is a larger version of the toy data above, but it is still a subset of the full data.

# library(readxl)
# note we are saving the data from the Excel file in our environment 
# with the object name yrbss_data
yrbss_data <- read_excel("data/yrbss_demo.xlsx")

First look at the data:

glimpse(yrbss_data)

Rows: 20,000
Columns: 8
$ record   <dbl> 931897, 333862, 36253, 1095530, 1303997, 261619, 926649, 1309…
$ age      <chr> "15 years old", "17 years old", "18 years old or older", "15 …
$ sex      <chr> "Female", "Female", "Male", "Male", "Male", "Male", "Male", "…
$ grade    <chr> "10th", "12th", "11th", "10th", "9th", "9th", "11th", "12th",…
$ race4    <chr> "White", "White", "Hispanic/Latino", "Black or African Americ…
$ race7    <chr> "White", "White", "Hispanic/Latino", "Black or African Americ…
$ bmi      <dbl> 17.1790, 20.2487, NA, 27.9935, 24.4922, NA, 20.5435, 19.2555,…
$ stweight <dbl> 54.43, 57.15, NA, 85.73, 66.68, NA, 70.31, 58.97, 123.38, NA,…

After loading the dataset, you will see yrbss_data in the Data list of the Environment window.

6.1 View the `YRBSS` dataset

Run the View command below within RStudio.
- A new tab in the scripting window will appear with the yrbss_data dataset.
Note that the code chunk specifies #| eval: FALSE so that this code is not run while rendering the .qmd file.

View(yrbss_data)

7 Summarizing data and `%>%`

Often we want to summarize columns of data
Quantitative data and categorical/text data will be summarized differently
We also introduce the pipe symbol %>%, which is used to string together commands

The pipe is read “and then” and tells R the next function will be applied to the data on the left side of the pipe:

yrbss_data %>%
  head() # shows top rows

# A tibble: 6 × 8
   record age                   sex    grade race4          race7   bmi stweight
    <dbl> <chr>                 <chr>  <chr> <chr>          <chr> <dbl>    <dbl>
1  931897 15 years old          Female 10th  White          White  17.2     54.4
2  333862 17 years old          Female 12th  White          White  20.2     57.2
3   36253 18 years old or older Male   11th  Hispanic/Lati… Hisp…  NA       NA  
4 1095530 15 years old          Male   10th  Black or Afri… Blac…  28.0     85.7
5 1303997 14 years old          Male   9th   All other rac… Mult…  24.5     66.7
6  261619 17 years old          Male   9th   All other rac… <NA>   NA       NA

8 Summarizing quantitative data

The rstatix package has a useful function called get_summary_stats() for quick statistics of quantitative data

# library(rstatix)

yrbss_data %>% 
  get_summary_stats(
    bmi, stweight,
    type = "common"
    )

# A tibble: 2 × 10
  variable     n   min   max median   iqr  mean    sd    se    ci
  <fct>    <dbl> <dbl> <dbl>  <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 bmi      13542  13.2  53.9   22.3  5.52  23.5  4.99 0.043 0.084
2 stweight 13443  27.7 181.    64.0 19.0   67.5 16.9  0.145 0.285

The type options are
- type = c("full", "common", "robust", "five_number", "mean_sd", "mean_se", "mean_ci", "median_iqr", "median_mad", "quantile", "mean", "median", "min", "max")

8.1 Stratified summaries (and more on `%>%`)

Now let’s summarize the BMI and weight variables by a strata subgroup.
We do this with the group_by() function
Notice that we now have two %>%:
- data %>% group_by(...) %>% get_summary_stats(...)
- With the pipe symbol %>% we can string together many commands in the order we want them applied. Example:
  - first tell R what dataset to use,
  - then have R group (stratify) the data by sex,
  - and then calculate summary statistics

yrbss_data %>% 
  group_by(sex) %>%
  get_summary_stats(
    bmi, stweight,
    type = "common"
    )

Warning: There was 1 warning in `mutate()`.
ℹ In argument: `sd = stats::sd(.value., na.rm = TRUE)`.
Caused by warning:
! There were 2 warnings in `dplyr::summarise()`.
The first warning was:
ℹ In argument: `min = min(.value., na.rm = TRUE)`.
Caused by warning in `min()`:
! no non-missing arguments to min; returning Inf
ℹ Run `dplyr::last_dplyr_warnings()` to see the 1 remaining warning.

# A tibble: 4 × 11
  sex    variable     n   min   max median   iqr  mean    sd    se    ci
  <chr>  <fct>    <dbl> <dbl> <dbl>  <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 Female bmi       6622  13.2  53.3   22.1  5.33  23.3  4.96 0.061 0.119
2 Female stweight  6542  27.7 181.    59.0 15.9   61.7 14.2  0.175 0.343
3 Male   bmi       6920  13.2  53.9   22.5  5.51  23.7  5.01 0.06  0.118
4 Male   stweight  6901  35.4 181.    69.4 20.0   73.1 17.4  0.209 0.41

8.2 Stratify by 2 variables

We can specify more than one variable within group_by() to stratify the summaries by more variables

yrbss_data %>% 
  group_by(sex, grade) %>%
  get_summary_stats(
    bmi, stweight,
    type = "common"
    )

Warning: There were 5 warnings in `mutate()`.
The first warning was:
ℹ In argument: `sd = stats::sd(.value., na.rm = TRUE)`.
Caused by warning:
! There were 2 warnings in `dplyr::summarise()`.
The first warning was:
ℹ In argument: `min = min(.value., na.rm = TRUE)`.
Caused by warning in `min()`:
! no non-missing arguments to min; returning Inf
ℹ Run `dplyr::last_dplyr_warnings()` to see the 1 remaining warning.
ℹ Run `dplyr::last_dplyr_warnings()` to see the 4 remaining warnings.

# A tibble: 20 × 12
   sex    grade variable     n   min   max median   iqr  mean    sd    se    ci
   <chr>  <chr> <fct>    <dbl> <dbl> <dbl>  <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
 1 Female 10th  bmi       1604  13.6  52.3   22.0  5.29  23.0  4.75 0.119 0.233
 2 Female 10th  stweight  1582  37.2 132.    58.1 15.0   61.1 13.2  0.332 0.652
 3 Female 11th  bmi       1640  13.6  52.4   22.3  5.41  23.4  4.78 0.118 0.232
 4 Female 11th  stweight  1624  36.3 161.    59.0 15.0   62.5 14.1  0.35  0.687
 5 Female 12th  bmi       1675  15.4  53.3   22.7  5.47  23.9  5.23 0.128 0.251
 6 Female 12th  stweight  1664  36.3 154.    59.9 15.4   63.4 14.8  0.363 0.713
 7 Female 9th   bmi       1655  13.2  51.2   21.7  5.26  22.8  4.95 0.122 0.239
 8 Female 9th   stweight  1625  27.7 181.    56.7 15.4   59.8 14.1  0.35  0.686
 9 Female <NA>  bmi         48  17.0  47.6   21.2  4.35  23.1  5.69 0.821 1.65 
10 Female <NA>  stweight    47  35.8 113.    56.2 13.2   60.3 14.4  2.10  4.23 
11 Male   10th  bmi       1739  13.7  51.5   22.4  5.38  23.5  4.77 0.114 0.224
12 Male   10th  stweight  1734  40.8 163.    69.0 18.1   72.3 16.5  0.397 0.778
13 Male   11th  bmi       1704  14.0  53.9   23.0  5.58  24.1  4.97 0.12  0.236
14 Male   11th  stweight  1704  40.8 172.    70.8 20.4   75.2 17.2  0.416 0.815
15 Male   12th  bmi       1611  14.4  52.3   23.1  5.79  24.5  5.15 0.128 0.252
16 Male   12th  stweight  1611  43.1 181.    72.6 20.4   76.9 17.9  0.446 0.876
17 Male   9th   bmi       1773  13.2  52.9   21.7  5.58  22.8  4.88 0.116 0.227
18 Male   9th   stweight  1761  35.8 181.    65.8 19.0   68.2 16.4  0.391 0.768
19 Male   <NA>  bmi         93  13.9  53.4   21.7  5.86  23.7  6.83 0.708 1.41 
20 Male   <NA>  stweight    91  35.4 167.    69.0 20.0   73.1 21.6  2.27  4.50

8.3 Prettier table of output

Below we add on gt() at the end to make the table prettier.
- Make sure to check this out in the html output.
The gt() command is from the gt package

yrbss_data %>% 
  group_by(sex, grade) %>%
  get_summary_stats(
    bmi, stweight,
    type = "common"
    ) %>%
  gt()

Warning: There were 5 warnings in `mutate()`.
The first warning was:
ℹ In argument: `sd = stats::sd(.value., na.rm = TRUE)`.
Caused by warning:
! There were 2 warnings in `dplyr::summarise()`.
The first warning was:
ℹ In argument: `min = min(.value., na.rm = TRUE)`.
Caused by warning in `min()`:
! no non-missing arguments to min; returning Inf
ℹ Run `dplyr::last_dplyr_warnings()` to see the 1 remaining warning.
ℹ Run `dplyr::last_dplyr_warnings()` to see the 4 remaining warnings.

sex	grade	variable	n	min	max	median	iqr	mean	sd	se	ci
Female	10th	bmi	1604	13.578	52.269	21.972	5.288	22.985	4.753	0.119	0.233
Female	10th	stweight	1582	37.200	131.540	58.060	14.970	61.071	13.216	0.332	0.652
Female	11th	bmi	1640	13.578	52.431	22.269	5.406	23.448	4.780	0.118	0.232
Female	11th	stweight	1624	36.290	160.570	58.970	14.970	62.536	14.116	0.350	0.687
Female	12th	bmi	1675	15.375	53.265	22.656	5.470	23.884	5.231	0.128	0.251
Female	12th	stweight	1664	36.290	154.220	59.880	15.420	63.434	14.822	0.363	0.713
Female	9th	bmi	1655	13.161	51.208	21.660	5.261	22.819	4.954	0.122	0.239
Female	9th	stweight	1625	27.670	180.990	56.700	15.420	59.802	14.103	0.350	0.686
Female	NA	bmi	48	16.991	47.610	21.216	4.349	23.092	5.687	0.821	1.651
Female	NA	stweight	47	35.830	113.400	56.250	13.155	60.261	14.413	2.102	4.232
Male	10th	bmi	1739	13.664	51.540	22.430	5.379	23.453	4.766	0.114	0.224
Male	10th	stweight	1734	40.820	163.300	68.950	18.140	72.318	16.513	0.397	0.778
Male	11th	bmi	1704	14.003	53.947	22.983	5.585	24.129	4.973	0.120	0.236
Male	11th	stweight	1704	40.820	172.370	70.760	20.420	75.238	17.158	0.416	0.815
Male	12th	bmi	1611	14.397	52.284	23.102	5.789	24.470	5.150	0.128	0.252
Male	12th	stweight	1611	43.090	180.990	72.580	20.410	76.900	17.918	0.446	0.876
Male	9th	bmi	1773	13.194	52.882	21.671	5.579	22.809	4.883	0.116	0.227
Male	9th	stweight	1761	35.830	180.990	65.770	19.050	68.180	16.426	0.391	0.768
Male	NA	bmi	93	13.896	53.432	21.698	5.862	23.744	6.827	0.708	1.406
Male	NA	stweight	91	35.380	167.380	68.950	19.955	73.075	21.634	2.268	4.505

8.4 Advanced examples using `summarize` function

Below are links to data summary examples in our previous OCTRI-BERD workshop on Introduction to R and RStudio for Exploratory Data Analysis: Part 2 using the function summarize():

Summarize examples: Slides 6 - 12
All slides:
- html slides: bit.ly/berd_intro_part2
- pdf slides: bit.ly/berd_intro_part2_pdf

9 Summarizing categorical data: frequency tables

We will be creating frequency tables using the tabyl function from the janitor package (loaded above).

9.1 Frequency table for one variable

Default table:

yrbss_data %>% 
  tabyl(age)

                     age    n percent valid_percent
 12 years old or younger  137 0.00685    0.00692094
            13 years old   96 0.00480    0.00484971
            14 years old 2026 0.10130    0.10234908
            15 years old 4290 0.21450    0.21672139
            16 years old 4924 0.24620    0.24874968
            17 years old 4988 0.24940    0.25198282
   18 years old or older 3334 0.16670    0.16842637
                    <NA>  205 0.01025            NA

Make the output prettier using adorn_ options and adding on gt():

With adornments:

yrbss_data %>% 
  tabyl(age)   %>%
  adorn_totals("row") %>% 
  adorn_pct_formatting(digits=2)

                     age     n percent valid_percent
 12 years old or younger   137   0.69%         0.69%
            13 years old    96   0.48%         0.48%
            14 years old  2026  10.13%        10.23%
            15 years old  4290  21.45%        21.67%
            16 years old  4924  24.62%        24.87%
            17 years old  4988  24.94%        25.20%
   18 years old or older  3334  16.67%        16.84%
                    <NA>   205   1.03%             -
                   Total 20000 100.00%       100.00%

With adornments & gt():

yrbss_data %>% 
  tabyl(age)   %>%
  adorn_totals("row") %>% 
  adorn_pct_formatting(digits=2) %>% 
  gt() # from the gt package

age	n	percent	valid_percent
12 years old or younger	137	0.69%	0.69%
13 years old	96	0.48%	0.48%
14 years old	2026	10.13%	10.23%
15 years old	4290	21.45%	21.67%
16 years old	4924	24.62%	24.87%
17 years old	4988	24.94%	25.20%
18 years old or older	3334	16.67%	16.84%
NA	205	1.03%	-
Total	20000	100.00%	100.00%

9.2 Contingency tables (two-way tables)

Contingency tables summarize data for two categorical variables
- with each value in the table representing the number of times
  a particular combination of outcomes occurs
Row & column totals added with adorn_totals option
- Can choose to include row and/or column totals

# ages vs. grades:

yrbss_data %>% 
  tabyl(age, grade) %>%    
  adorn_totals(c("row", "col")) %>% 
  adorn_title(placement = "combined") %>% 
  gt()

age/grade	10th	11th	12th	9th	NA_	Total
12 years old or younger	9	21	35	31	41	137
13 years old	6	2	1	79	8	96
14 years old	65	8	6	1922	25	2026
15 years old	1909	104	22	2225	30	4290
16 years old	2102	1899	157	717	49	4924
17 years old	637	2178	1956	164	53	4988
18 years old or older	163	670	2373	60	68	3334
NA	16	9	27	21	132	205
Total	4907	4891	4577	5219	406	20000

9.3 Contingency tables with percentages

How are the percentages calculated in the code below?
What else was added to the table, other than the percentages?

yrbss_data %>% 
  tabyl(age, grade) %>%    
  adorn_totals(c("row")) %>%
  adorn_percentages("row") %>%   
  adorn_pct_formatting(digits=0) %>%    
  adorn_ns() %>% 
  gt()

age	10th	11th	12th	9th	NA_
12 years old or younger	7% (9)	15% (21)	26% (35)	23% (31)	30% (41)
13 years old	6% (6)	2% (2)	1% (1)	82% (79)	8% (8)
14 years old	3% (65)	0% (8)	0% (6)	95% (1,922)	1% (25)
15 years old	44% (1,909)	2% (104)	1% (22)	52% (2,225)	1% (30)
16 years old	43% (2,102)	39% (1,899)	3% (157)	15% (717)	1% (49)
17 years old	13% (637)	44% (2,178)	39% (1,956)	3% (164)	1% (53)
18 years old or older	5% (163)	20% (670)	71% (2,373)	2% (60)	2% (68)
NA	8% (16)	4% (9)	13% (27)	10% (21)	64% (132)
Total	25% (4,907)	24% (4,891)	23% (4,577)	26% (5,219)	2% (406)

9.4 Stratified (3-way) tables

We can also add a third variable, which creates 2x2 tables stratified by the third variable:

yrbss_data %>% 
  tabyl(age, grade, sex)

$Female
                     age 10th 11th 12th  9th NA_
 12 years old or younger    3    5   14   15  21
            13 years old    0    2    0   40   1
            14 years old   30    4    2  963   8
            15 years old  961   53   11 1070   9
            16 years old 1016  957   81  311  13
            17 years old  261 1055 1044   66  21
   18 years old or older   57  285 1114   19  29
                    <NA>    4    4   11    8  24

$Male
                     age 10th 11th 12th  9th NA_
 12 years old or younger    6   16   20   16  18
            13 years old    4    0    1   38   6
            14 years old   34    4    4  950  14
            15 years old  942   51   10 1136  20
            16 years old 1067  936   76  395  32
            17 years old  369 1106  896   98  30
   18 years old or older  105  379 1242   40  35
                    <NA>   12    4   14   11  40

$NA_
                     age 10th 11th 12th 9th NA_
 12 years old or younger    0    0    1   0   2
            13 years old    2    0    0   1   1
            14 years old    1    0    0   9   3
            15 years old    6    0    1  19   1
            16 years old   19    6    0  11   4
            17 years old    7   17   16   0   2
   18 years old or older    1    6   17   1   4
                    <NA>    0    1    2   2  68

Make sure to scroll down in the output box above to see all of the output

9.5 More `tabyl` and `adorn_` options

See the tabyls vignette for more tabyl and adorn_ options:

https://cran.r-project.org/web/packages/janitor/vignettes/tabyls.html

10 Practice time! (part 2)

Create a contingency table for race4 and race7, with both percentages and counts. What is the difference between these two variables?

Create a contingency table for race4 and race7, with both percentages and counts, stratified by grade.

Calculate the minimum, median, and maximum BMI and weights of participants in the YRBSS data.

Calculate the minimum, median, and maximum BMI and weights of the participants in the YRBSS data, now stratified by age.

11 Data visualization

This is a very brief introduction to creating plots using the ggplot2 package.
- The ggplot2 package gets loaded with the tidyverse and thus we do not need to load it separately.

12 Visualizing quantitative variables

What is the distribution shape of a quantitative variable?
Does the distribution shape of a quantitative variable change when stratified by a categorical variable?
Is there an association between two quantitative variables?

12.1 Histograms

ggplot(data = yrbss_data,    # specify the dataset
       # aes = aesthetics: specify variables being used
       aes(x = bmi)) +  
  geom_histogram() # type of plot

`stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

Warning: Removed 6458 rows containing non-finite outside the scale range
(`stat_bin()`).

12.2 Boxplots

Recall that the whiskers’ ends are the minimum and maximum values
The box is from the first to third quartiles, with the line inside the box at the median

BMI: horizontal boxplot (specify x = ...)

ggplot(data = yrbss_data,   # dataset
  # aes: specify vars being used
       aes(x = bmi)) +   
  geom_boxplot()   # type of plot

Warning: Removed 6458 rows containing non-finite outside the scale range
(`stat_boxplot()`).

Weight: vertical boxplot (specify y = ...)

ggplot(data = yrbss_data, 
# use y instead of x in aes for vertical:
  aes(y = stweight)) + 
  geom_boxplot()

Warning: Removed 6557 rows containing non-finite outside the scale range
(`stat_boxplot()`).

12.3 Stratified boxplots (plus axis labels & title!)

ggplot(data = yrbss_data, 
       aes(
         # horizontal boxplots for BMI (x=):
         x = bmi,
         # stratified vertically (y=) by age:
         y = age
         )) + 
  geom_boxplot() + 
  # Add on labels and a title:
  labs(x = "BMI", 
       y = "Age", 
       title =  "BMI distributions by age category"
       )

Warning: Removed 6458 rows containing non-finite outside the scale range
(`stat_boxplot()`).

12.4 Add color using `fill`

Also remove the NA category with drop_na() (we will talk about this in later slides)
What’s the difference in use of fill= here?

ggplot(
  data = yrbss_data %>% drop_na(age), #NEW!
  aes(
    x = bmi,
    y = age,
    fill = age  # NEW!
  )) + 
  geom_boxplot() +
  # remove legend:
  theme(legend.position="none")

Warning: Removed 6253 rows containing non-finite outside the scale range
(`stat_boxplot()`).

ggplot(
  data = yrbss_data %>% drop_na(age), #NEW!
  aes(
    x = bmi,
    y = age,
    fill = sex  # NEW!
  )) + 
  geom_boxplot()

Warning: Removed 6253 rows containing non-finite outside the scale range
(`stat_boxplot()`).

12.5 Scatterplots

ggplot(data = yrbss_data, # specify the dataset
       # aes = aesthetics: specify variables being used
       aes(x = stweight,       
           y = bmi
           )) +  
  geom_point()    # type of plot

Warning: Removed 6559 rows containing missing values or values outside the scale range
(`geom_point()`).

12.6 Scatterplot with color-coded dots

ggplot(data = yrbss_data, 
       # aes = aesthetics: specify variables being used
       aes(x = stweight,      
           y = bmi,  
           color = sex,
           )) +   
  geom_point(alpha = .2) # alpha controls transparency of the points

Warning: Removed 6559 rows containing missing values or values outside the scale range
(`geom_point()`).

12.7 Color vs. fill vs. alpha

color (or colour) is used for lines and outlines
fill is used for interiors
Note: geom_point() shape default is a point with only an outline and no interior -> we used color and not fill
alpha specifies the opacity / transparency
- Values range from 0 to 1
- lower alphas are more transparent
See ggplot webpage Colour related aesthetics: colour, fill, and alpha for more information

13 Visualizing categorical variables

13.1 Barplots

By default, barplots display counts (frequencies) on the vertical axis:

ggplot(data = yrbss_data, 
       aes(x = age)) +
  geom_bar() + # plot type 
  # angle the x-axis labels so that they are not overlapping
  theme(axis.text.x = element_text(angle = -45, hjust = 0))

13.2 Barplots with proportions

To show proportions, the code is more complicated.

ggplot(data = yrbss_data, 
       aes(x = age)) +
  # specify aesthetics within the barplot to show proportions
  geom_bar(aes(y = after_stat(prop), group = 1)) + 
  
  # Next line converts y-axis labels to percentages instead of proportions
  scale_y_continuous(labels =  scales::percent_format()) +
  
  # "dodge" the x-axis labels
  scale_x_discrete(guide = guide_axis(n.dodge = 2))

13.3 Barplots with 2 variables: segmented bar plots

Stacked bars showing counts:

ggplot(data = yrbss_data, 
       aes(x = age,
           # specify variable for colors 
           fill = grade)) + 
  geom_bar()

For proportions within each bar:
(added position = "fill" within geom_bar())

ggplot(data = yrbss_data, 
       aes(x = age,
           # specify variable for colors
           fill = grade)) + 
  geom_bar(position = "fill")

13.4 Side-by-side barplots

ggplot(data = yrbss_data, 
       aes(x = age,
           # specify variable for colors 
           fill = grade)) + 
  geom_bar(position = "dodge") +
  # angle the x-axis labels so that they are not overlapping
  theme(axis.text.x = element_text(angle = -15, hjust = 0))

13.5 Saving a ggplot as an image file

First, create a plot and save it as an R object:

myplot <- ggplot(data = yrbss_data, 
       aes(x = age,
           # specify variable for colors 
           fill = grade)) + 
  geom_bar() +
  # angle the x-axis labels so that they are not overlapping
  theme(axis.text.x = element_text(angle = -15, hjust = 0))

myplot

Save the plot as a pdf (or “jpeg”, “tiff”, “png”, etc.)

ggsave(
  filename = "output/myplot.pdf",
  plot = myplot,
  device = "pdf",
  width = 5,
  height = 3,
  units = "in",
  dpi = 200)

Can specify the dpi when saving, many journals have dpi requirements.

See ggsave webpage for more details: https://ggplot2.tidyverse.org/reference/ggsave.html

14 Missing data summary & visualization

I highly recommend the tutorial: Exploring missing values in naniar https://allisonhorst.shinyapps.io/missingexplorer/
These slides have some highlights of tools presented in the tutorial

Load the naniar package if haven’t already:

library(naniar)

14.1 Missingness statistics

Total number of missing values in dataset:

n_miss(yrbss_data)

[1] 15754

Proportion of missing values in dataset:

prop_miss(yrbss_data)

[1] 0.0984625

Missingness by each variable:

miss_var_summary(yrbss_data) %>% gt()

variable	n_miss	pct_miss
stweight	6557	32.8
bmi	6458	32.3
race7	1187	5.94
race4	710	3.55
grade	406	2.03
sex	231	1.16
age	205	1.03
record	0	0

14.2 Visualizing missing data across dataset

This plot helps visualize whether
- there are variables with a lot of missing values and/or
- participants with a lot of missing values
Each row is an observation
Columns are for the different variables in the dataset
Black = missing, grey = not missing

vis_miss(yrbss_data)

14.3 Missingness heatmap by group

This plot is useful in comparing missingness rates stratified by a variable
Specify the variable to stratify the missingness heatmap by in the fct = option

Example: Percent missingness stratified by age group

gg_miss_fct(yrbss_data, fct = age)

14.4 Missingness intersections (UpSet plot)

UpSet plots are great for finding sets of variables that have a lot of missingess.

gg_miss_upset(yrbss_data, nsets = 10)

15 Practice time! (part 3)

Create a histogram for the stweight variable. Describe the distribution shape.

What happens if you change x = to y = in the histogram code?

Create boxplots of stweight stratified by sex.

Create a barplot of race4 categories with the y axis showing percentages.

Visualize the missing data across the YRBSS subset dataset we created toy_data using vis_miss(). What missingness patterns do you see?

16 Data wrangling

16.1 Sorting data frames using `arrange()`

arrange() is a function that lets us sort a data.frame by a specified variable.

By default, it sorts in ascending order:

yrbss_data %>% 
  arrange(record)

# A tibble: 20,000 × 8
   record age                     sex   grade race4         race7   bmi stweight
    <dbl> <chr>                   <chr> <chr> <chr>         <chr> <dbl>    <dbl>
 1  30592 <NA>                    <NA>  <NA>  <NA>          <NA>     NA       NA
 2  30593 <NA>                    <NA>  9th   Hispanic/Lat… Hisp…    NA       NA
 3  30595 16 years old            <NA>  10th  White         White    NA       NA
 4  30599 12 years old or younger Male  <NA>  Hispanic/Lat… Hisp…    NA       NA
 5  30601 12 years old or younger Male  11th  White         White    NA       NA
 6  30604 12 years old or younger Male  9th   Hispanic/Lat… Hisp…    NA       NA
 7  30605 12 years old or younger Male  9th   All other ra… Asian    NA       NA
 8  30607 12 years old or younger Male  10th  All other ra… Am I…    NA       NA
 9  30608 12 years old or younger Male  9th   All other ra… <NA>     NA       NA
10  30612 14 years old            Male  9th   All other ra… <NA>     NA       NA
# ℹ 19,990 more rows

To sort by descending order, you need to wrap the variable in the desc() function:

yrbss_data %>% 
  arrange(desc(record))

# A tibble: 20,000 × 8
    record age                   sex    grade race4         race7   bmi stweight
     <dbl> <chr>                 <chr>  <chr> <chr>         <chr> <dbl>    <dbl>
 1 1317069 18 years old or older Female 12th  Black or Afr… Blac…  24.4     69.0
 2 1317067 18 years old or older Female 12th  White         White  18.8     49.9
 3 1317066 18 years old or older Female 12th  White         White  32.1     90.7
 4 1317065 18 years old or older Female 12th  White         White  21.2     56.2
 5 1317064 18 years old or older Female 12th  White         White  23.5     68.0
 6 1317063 18 years old or older Female 12th  White         White  21.7     52.2
 7 1317062 18 years old or older Female 12th  White         White  24.8     61.2
 8 1317058 18 years old or older Female 11th  Black or Afr… Blac…  17.7     49.9
 9 1317057 18 years old or older Female 12th  All other ra… Asian  19.8     44.4
10 1317056 18 years old or older Female 12th  White         White  21.7     61.2
# ℹ 19,990 more rows

16.2 Sorting data frames by multiple variables

You can also arrange by multiple variables.
- The code below sorts by age first, and then within age, we sort by record.

yrbss_data %>% 
  arrange(age, record)

# A tibble: 20,000 × 8
   record age                     sex    grade race4        race7   bmi stweight
    <dbl> <chr>                   <chr>  <chr> <chr>        <chr> <dbl>    <dbl>
 1  30599 12 years old or younger Male   <NA>  Hispanic/La… Hisp…    NA       NA
 2  30601 12 years old or younger Male   11th  White        White    NA       NA
 3  30604 12 years old or younger Male   9th   Hispanic/La… Hisp…    NA       NA
 4  30605 12 years old or younger Male   9th   All other r… Asian    NA       NA
 5  30607 12 years old or younger Male   10th  All other r… Am I…    NA       NA
 6  30608 12 years old or younger Male   9th   All other r… <NA>     NA       NA
 7  36582 12 years old or younger Female 12th  All other r… <NA>     NA       NA
 8  36584 12 years old or younger Female <NA>  Black or Af… Blac…    NA       NA
 9  36585 12 years old or younger Female 12th  All other r… Am I…    NA       NA
10  36586 12 years old or younger Female 9th   Black or Af… Blac…    NA       NA
# ℹ 19,990 more rows

Note that order of variables in arrange() matters!

yrbss_data %>% 
  arrange(record, age)

# A tibble: 20,000 × 8
   record age                     sex   grade race4         race7   bmi stweight
    <dbl> <chr>                   <chr> <chr> <chr>         <chr> <dbl>    <dbl>
 1  30592 <NA>                    <NA>  <NA>  <NA>          <NA>     NA       NA
 2  30593 <NA>                    <NA>  9th   Hispanic/Lat… Hisp…    NA       NA
 3  30595 16 years old            <NA>  10th  White         White    NA       NA
 4  30599 12 years old or younger Male  <NA>  Hispanic/Lat… Hisp…    NA       NA
 5  30601 12 years old or younger Male  11th  White         White    NA       NA
 6  30604 12 years old or younger Male  9th   Hispanic/Lat… Hisp…    NA       NA
 7  30605 12 years old or younger Male  9th   All other ra… Asian    NA       NA
 8  30607 12 years old or younger Male  10th  All other ra… Am I…    NA       NA
 9  30608 12 years old or younger Male  9th   All other ra… <NA>     NA       NA
10  30612 14 years old            Male  9th   All other ra… <NA>     NA       NA
# ℹ 19,990 more rows

16.3 `arrange()` after `count()`-ing categorical data

Another example of when arrange() is useful
We’ve used tabyl() to see what values exist in a categorical variable.
We can also use count() to count all of the unique values for a categorical variable.
This will also show NA category, which is the special missing variable type:

yrbss_data %>%
  count(age)

# A tibble: 8 × 2
  age                         n
  <chr>                   <int>
1 12 years old or younger   137
2 13 years old               96
3 14 years old             2026
4 15 years old             4290
5 16 years old             4924
6 17 years old             4988
7 18 years old or older    3334
8 <NA>                      205

Sometimes it helps to arrange by that category if you have many categories to view.
This can also alert us if there are categorical coding mistakes (such as misspellings) in our data.

yrbss_data %>%
  count(age) %>%
  arrange(n)

# A tibble: 8 × 2
  age                         n
  <chr>                   <int>
1 13 years old               96
2 12 years old or younger   137
3 <NA>                      205
4 14 years old             2026
5 18 years old or older    3334
6 15 years old             4290
7 16 years old             4924
8 17 years old             4988

16.4 Saving changes we make to a dataframe

Note that we have not changed the order of the yrbss_data dataframe after arranging:

head(yrbss_data)

# A tibble: 6 × 8
   record age                   sex    grade race4          race7   bmi stweight
    <dbl> <chr>                 <chr>  <chr> <chr>          <chr> <dbl>    <dbl>
1  931897 15 years old          Female 10th  White          White  17.2     54.4
2  333862 17 years old          Female 12th  White          White  20.2     57.2
3   36253 18 years old or older Male   11th  Hispanic/Lati… Hisp…  NA       NA  
4 1095530 15 years old          Male   10th  Black or Afri… Blac…  28.0     85.7
5 1303997 14 years old          Male   9th   All other rac… Mult…  24.5     66.7
6  261619 17 years old          Male   9th   All other rac… <NA>   NA       NA

We need to save our work by re-assigning the sorted/arranged data frame as yrbss_data with the <- operator:

yrbss_data <- yrbss_data %>% arrange(desc(record))

yrbss_data

# A tibble: 20,000 × 8
    record age                   sex    grade race4         race7   bmi stweight
     <dbl> <chr>                 <chr>  <chr> <chr>         <chr> <dbl>    <dbl>
 1 1317069 18 years old or older Female 12th  Black or Afr… Blac…  24.4     69.0
 2 1317067 18 years old or older Female 12th  White         White  18.8     49.9
 3 1317066 18 years old or older Female 12th  White         White  32.1     90.7
 4 1317065 18 years old or older Female 12th  White         White  21.2     56.2
 5 1317064 18 years old or older Female 12th  White         White  23.5     68.0
 6 1317063 18 years old or older Female 12th  White         White  21.7     52.2
 7 1317062 18 years old or older Female 12th  White         White  24.8     61.2
 8 1317058 18 years old or older Female 11th  Black or Afr… Blac…  17.7     49.9
 9 1317057 18 years old or older Female 12th  All other ra… Asian  19.8     44.4
10 1317056 18 years old or older Female 12th  White         White  21.7     61.2
# ℹ 19,990 more rows

17 Subsetting data

18 Subset rows: `filter()`ing data

18.1 `filter()`ing quantitative data

filter() is an extremely powerful function. It lets us subset our data according to specific criteria.

Let’s filter on a numeric variable, bmi:

Note the double equal signs! == is a test of whether two things are equal, it is not an assignment (i.e. x = 5 assigns the value 5 to the object x).

yrbss_data %>%
  filter(bmi > 40)

# A tibble: 160 × 8
    record age                   sex    grade race4         race7   bmi stweight
     <dbl> <chr>                 <chr>  <chr> <chr>         <chr> <dbl>    <dbl>
 1 1316985 18 years old or older Female 12th  White         White  42.4    127. 
 2 1316823 18 years old or older Female 12th  Black or Afr… Blac…  43.3    118. 
 3 1316755 18 years old or older Female 12th  White         White  41.0    109. 
 4 1316476 18 years old or older Female 12th  Hispanic/Lat… Hisp…  49.6    104. 
 5 1316448 18 years old or older Female 12th  All other ra… Mult…  41.9    128. 
 6 1316336 18 years old or older Female 12th  Black or Afr… Blac…  40.2    113. 
 7 1316330 18 years old or older Female 12th  Hispanic/Lat… Hisp…  40.2    113. 
 8 1316235 18 years old or older Female 12th  <NA>          <NA>   45.1    120. 
 9 1316186 18 years old or older Female <NA>  Hispanic/Lat… Hisp…  47.6     89.4
10 1315913 17 years old          Female 12th  Black or Afr… Blac…  53.3    142. 
# ℹ 150 more rows

yrbss_data %>%
  filter(bmi == 13.9)

# A tibble: 0 × 8
# ℹ 8 variables: record <dbl>, age <chr>, sex <chr>, grade <chr>, race4 <chr>,
#   race7 <chr>, bmi <dbl>, stweight <dbl>

yrbss_data %>%
  filter(bmi == 13.9229)

# A tibble: 1 × 8
   record age          sex   grade race4                    race7   bmi stweight
    <dbl> <chr>        <chr> <chr> <chr>                    <chr> <dbl>    <dbl>
1 1096565 15 years old Male  10th  Black or African Americ… Blac…  13.9     56.2

18.2 `filter()`ing cateogrical data

To filter on a categorical variable, we need to know the values (levels) of the variable:

yrbss_data %>% tabyl(grade)

 grade    n percent valid_percent
  10th 4907 0.24535     0.2504338
  11th 4891 0.24455     0.2496172
  12th 4577 0.22885     0.2335919
   9th 5219 0.26095     0.2663570
  <NA>  406 0.02030            NA

Note the double equal signs! == is a test of whether two things are equal, it is not an assignment (i.e. x = 5 assigns the value 5 to the object x).

yrbss_data %>%
  filter(grade == "10th") #We must include " " around the value for categorical variables

# A tibble: 4,907 × 8
    record age                   sex    grade race4         race7   bmi stweight
     <dbl> <chr>                 <chr>  <chr> <chr>         <chr> <dbl>    <dbl>
 1 1317010 18 years old or older Female 10th  Black or Afr… Blac…  28.3     61.2
 2 1315996 17 years old          Female 10th  White         White  16.3     44.4
 3 1315971 17 years old          Female 10th  Hispanic/Lat… Hisp…  19.6     56.7
 4 1315934 17 years old          Female 10th  Black or Afr… Blac…  23.5     70.3
 5 1315928 17 years old          Female 10th  Black or Afr… Blac…  21.0     45.4
 6 1315926 17 years old          Female 10th  Hispanic/Lat… Hisp…  25.9     66.2
 7 1315884 17 years old          Female 10th  White         White  51.4    132. 
 8 1315816 17 years old          Female 10th  All other ra… Mult…  22.2     59.0
 9 1315792 17 years old          Female 10th  Hispanic/Lat… Hisp…  21.7     61.2
10 1315776 17 years old          Female 10th  Black or Afr… Blac…  20.6     50.8
# ℹ 4,897 more rows

If we want everything except 10th grade, we use the “not equal” sign !=

yrbss_data %>%
  filter(grade != "10th")

# A tibble: 14,687 × 8
    record age                   sex    grade race4         race7   bmi stweight
     <dbl> <chr>                 <chr>  <chr> <chr>         <chr> <dbl>    <dbl>
 1 1317069 18 years old or older Female 12th  Black or Afr… Blac…  24.4     69.0
 2 1317067 18 years old or older Female 12th  White         White  18.8     49.9
 3 1317066 18 years old or older Female 12th  White         White  32.1     90.7
 4 1317065 18 years old or older Female 12th  White         White  21.2     56.2
 5 1317064 18 years old or older Female 12th  White         White  23.5     68.0
 6 1317063 18 years old or older Female 12th  White         White  21.7     52.2
 7 1317062 18 years old or older Female 12th  White         White  24.8     61.2
 8 1317058 18 years old or older Female 11th  Black or Afr… Blac…  17.7     49.9
 9 1317057 18 years old or older Female 12th  All other ra… Asian  19.8     44.4
10 1317056 18 years old or older Female 12th  White         White  21.7     61.2
# ℹ 14,677 more rows

18.3 Filtering requires a little logic

We can chain multiple criteria using the & (AND) or | (OR) operators.
- But we need to review a little logic before we do this.
If we wanted to return participants who identified as
- Female and are in 10th grade,
- we would use an & to chain these criteria together:

sex == "Female" & grade == "10th"

Note that SPELLING AND CASE MATTER!
- You must write the name of the column exactly and
- also the value (i.e. “Female” not “female”) exactly.

18.4 Filtering with `&` (AND) criteria

yrbss_data %>%
  filter(sex == "Female" & 
           grade == "10th")

# A tibble: 2,332 × 8
    record age                   sex    grade race4         race7   bmi stweight
     <dbl> <chr>                 <chr>  <chr> <chr>         <chr> <dbl>    <dbl>
 1 1317010 18 years old or older Female 10th  Black or Afr… Blac…  28.3     61.2
 2 1315996 17 years old          Female 10th  White         White  16.3     44.4
 3 1315971 17 years old          Female 10th  Hispanic/Lat… Hisp…  19.6     56.7
 4 1315934 17 years old          Female 10th  Black or Afr… Blac…  23.5     70.3
 5 1315928 17 years old          Female 10th  Black or Afr… Blac…  21.0     45.4
 6 1315926 17 years old          Female 10th  Hispanic/Lat… Hisp…  25.9     66.2
 7 1315884 17 years old          Female 10th  White         White  51.4    132. 
 8 1315816 17 years old          Female 10th  All other ra… Mult…  22.2     59.0
 9 1315792 17 years old          Female 10th  Hispanic/Lat… Hisp…  21.7     61.2
10 1315776 17 years old          Female 10th  Black or Afr… Blac…  20.6     50.8
# ℹ 2,322 more rows

Note that we could also use the comma in this case:

yrbss_data %>%
  filter(sex == "Female",    # comma is equivalent to & within filter()
         grade == "10th")

# A tibble: 2,332 × 8
    record age                   sex    grade race4         race7   bmi stweight
     <dbl> <chr>                 <chr>  <chr> <chr>         <chr> <dbl>    <dbl>
 1 1317010 18 years old or older Female 10th  Black or Afr… Blac…  28.3     61.2
 2 1315996 17 years old          Female 10th  White         White  16.3     44.4
 3 1315971 17 years old          Female 10th  Hispanic/Lat… Hisp…  19.6     56.7
 4 1315934 17 years old          Female 10th  Black or Afr… Blac…  23.5     70.3
 5 1315928 17 years old          Female 10th  Black or Afr… Blac…  21.0     45.4
 6 1315926 17 years old          Female 10th  Hispanic/Lat… Hisp…  25.9     66.2
 7 1315884 17 years old          Female 10th  White         White  51.4    132. 
 8 1315816 17 years old          Female 10th  All other ra… Mult…  22.2     59.0
 9 1315792 17 years old          Female 10th  Hispanic/Lat… Hisp…  21.7     61.2
10 1315776 17 years old          Female 10th  Black or Afr… Blac…  20.6     50.8
# ℹ 2,322 more rows

18.5 Filtering with `|` (OR) criteria

If we wanted patients who were

Female or in 10th grade

we would use a | to chain these criteria together.

yrbss_data %>%
  filter(sex == "Female" |
           grade == "10th")

# A tibble: 12,167 × 8
    record age                   sex    grade race4         race7   bmi stweight
     <dbl> <chr>                 <chr>  <chr> <chr>         <chr> <dbl>    <dbl>
 1 1317069 18 years old or older Female 12th  Black or Afr… Blac…  24.4     69.0
 2 1317067 18 years old or older Female 12th  White         White  18.8     49.9
 3 1317066 18 years old or older Female 12th  White         White  32.1     90.7
 4 1317065 18 years old or older Female 12th  White         White  21.2     56.2
 5 1317064 18 years old or older Female 12th  White         White  23.5     68.0
 6 1317063 18 years old or older Female 12th  White         White  21.7     52.2
 7 1317062 18 years old or older Female 12th  White         White  24.8     61.2
 8 1317058 18 years old or older Female 11th  Black or Afr… Blac…  17.7     49.9
 9 1317057 18 years old or older Female 12th  All other ra… Asian  19.8     44.4
10 1317056 18 years old or older Female 12th  White         White  21.7     61.2
# ℹ 12,157 more rows

Think about it:
- Which of the AND vs. OR code blocks will return a larger number of participants?

18.6 Filtering out missing values using `drop_na()` (1/2)

We may want to remove rows that have missing data, which are coded as NA.
See the drop_na() reference for examples (?drop_na).
First look at rates of missingness using the naniar package:

naniar::miss_var_summary(yrbss_data)

# A tibble: 8 × 3
  variable n_miss pct_miss
  <chr>     <int>    <num>
1 stweight   6557    32.8 
2 bmi        6458    32.3 
3 race7      1187     5.94
4 race4       710     3.55
5 grade       406     2.03
6 sex         231     1.16
7 age         205     1.03
8 record        0     0

18.7 Filtering out missing values using `drop_na()` (2/2)

Pay attention to the number of rows for each of these results

Remove missingness in one column:

yrbss_data %>% drop_na(race7)

# A tibble: 18,813 × 8
    record age                   sex    grade race4         race7   bmi stweight
     <dbl> <chr>                 <chr>  <chr> <chr>         <chr> <dbl>    <dbl>
 1 1317069 18 years old or older Female 12th  Black or Afr… Blac…  24.4     69.0
 2 1317067 18 years old or older Female 12th  White         White  18.8     49.9
 3 1317066 18 years old or older Female 12th  White         White  32.1     90.7
 4 1317065 18 years old or older Female 12th  White         White  21.2     56.2
 5 1317064 18 years old or older Female 12th  White         White  23.5     68.0
 6 1317063 18 years old or older Female 12th  White         White  21.7     52.2
 7 1317062 18 years old or older Female 12th  White         White  24.8     61.2
 8 1317058 18 years old or older Female 11th  Black or Afr… Blac…  17.7     49.9
 9 1317057 18 years old or older Female 12th  All other ra… Asian  19.8     44.4
10 1317056 18 years old or older Female 12th  White         White  21.7     61.2
# ℹ 18,803 more rows

Remove missingness in two columns:

yrbss_data %>% drop_na(race7, bmi)

# A tibble: 13,126 × 8
    record age                   sex    grade race4         race7   bmi stweight
     <dbl> <chr>                 <chr>  <chr> <chr>         <chr> <dbl>    <dbl>
 1 1317069 18 years old or older Female 12th  Black or Afr… Blac…  24.4     69.0
 2 1317067 18 years old or older Female 12th  White         White  18.8     49.9
 3 1317066 18 years old or older Female 12th  White         White  32.1     90.7
 4 1317065 18 years old or older Female 12th  White         White  21.2     56.2
 5 1317064 18 years old or older Female 12th  White         White  23.5     68.0
 6 1317063 18 years old or older Female 12th  White         White  21.7     52.2
 7 1317062 18 years old or older Female 12th  White         White  24.8     61.2
 8 1317058 18 years old or older Female 11th  Black or Afr… Blac…  17.7     49.9
 9 1317057 18 years old or older Female 12th  All other ra… Asian  19.8     44.4
10 1317056 18 years old or older Female 12th  White         White  21.7     61.2
# ℹ 13,116 more rows

Remove all missingness:

yrbss_data %>% drop_na()

# A tibble: 12,897 × 8
    record age                   sex    grade race4         race7   bmi stweight
     <dbl> <chr>                 <chr>  <chr> <chr>         <chr> <dbl>    <dbl>
 1 1317069 18 years old or older Female 12th  Black or Afr… Blac…  24.4     69.0
 2 1317067 18 years old or older Female 12th  White         White  18.8     49.9
 3 1317066 18 years old or older Female 12th  White         White  32.1     90.7
 4 1317065 18 years old or older Female 12th  White         White  21.2     56.2
 5 1317064 18 years old or older Female 12th  White         White  23.5     68.0
 6 1317063 18 years old or older Female 12th  White         White  21.7     52.2
 7 1317062 18 years old or older Female 12th  White         White  24.8     61.2
 8 1317058 18 years old or older Female 11th  Black or Afr… Blac…  17.7     49.9
 9 1317057 18 years old or older Female 12th  All other ra… Asian  19.8     44.4
10 1317056 18 years old or older Female 12th  White         White  21.7     61.2
# ℹ 12,887 more rows

18.8 A warning about `filter()`: check for NA

Suppose we want to remove the 10th grade category.
However this column has missingness in it as we can see with tabyl().
When we try to filter using !="10th" we lose our NAs!
- That isn’t really expected.
You need to account for this with an or | statement:

yrbss_data %>% 
  tabyl(grade)

 grade    n percent valid_percent
  10th 4907 0.24535     0.2504338
  11th 4891 0.24455     0.2496172
  12th 4577 0.22885     0.2335919
   9th 5219 0.26095     0.2663570
  <NA>  406 0.02030            NA

Lost the NA’s in the filtering below!

yrbss_data %>% 
  filter(grade != "10th") %>%
  tabyl(grade)

 grade    n   percent
  11th 4891 0.3330156
  12th 4577 0.3116361
   9th 5219 0.3553483

Updated code to keep the NA’s when filtering:

yrbss_data %>%
  # add an | (OR) statement 
  # to keep in the NA values
  filter((grade != "10th") |
           is.na(grade)) %>%
  tabyl(grade)

 grade    n    percent valid_percent
  11th 4891 0.32405751     0.3330156
  12th 4577 0.30325316     0.3116361
   9th 5219 0.34578944     0.3553483
  <NA>  406 0.02689989            NA

19 Subset columns: `select()`ing data

19.1 Selecting columns using `select()`

The select() function allows us to select variables or columns from our dataset:

yrbss_data %>%
  select(record, age, race4)

# A tibble: 20,000 × 3
    record age                   race4                    
     <dbl> <chr>                 <chr>                    
 1 1317069 18 years old or older Black or African American
 2 1317067 18 years old or older White                    
 3 1317066 18 years old or older White                    
 4 1317065 18 years old or older White                    
 5 1317064 18 years old or older White                    
 6 1317063 18 years old or older White                    
 7 1317062 18 years old or older White                    
 8 1317058 18 years old or older Black or African American
 9 1317057 18 years old or older All other races          
10 1317056 18 years old or older White                    
# ℹ 19,990 more rows

If we want to select everything except one variable,
- we can use a - in front of that variable.

yrbss_data %>%
  select(-race4)

# A tibble: 20,000 × 7
    record age                   sex    grade race7                 bmi stweight
     <dbl> <chr>                 <chr>  <chr> <chr>               <dbl>    <dbl>
 1 1317069 18 years old or older Female 12th  Black or African A…  24.4     69.0
 2 1317067 18 years old or older Female 12th  White                18.8     49.9
 3 1317066 18 years old or older Female 12th  White                32.1     90.7
 4 1317065 18 years old or older Female 12th  White                21.2     56.2
 5 1317064 18 years old or older Female 12th  White                23.5     68.0
 6 1317063 18 years old or older Female 12th  White                21.7     52.2
 7 1317062 18 years old or older Female 12th  White                24.8     61.2
 8 1317058 18 years old or older Female 11th  Black or African A…  17.7     49.9
 9 1317057 18 years old or older Female 12th  Asian                19.8     44.4
10 1317056 18 years old or older Female 12th  White                21.7     61.2
# ℹ 19,990 more rows

19.2 `tidyselect`

There are ways to search column names and to select them.
These are called the tidyselect helpers.
You can see some examples here: https://tidyselect.r-lib.org/reference/language.html
For instance, you can see how this might be useful, where we select all columns where the column name includes the word “race”:

yrbss_data %>% 
  select(contains("race"))

# A tibble: 20,000 × 2
   race4                     race7                    
   <chr>                     <chr>                    
 1 Black or African American Black or African American
 2 White                     White                    
 3 White                     White                    
 4 White                     White                    
 5 White                     White                    
 6 White                     White                    
 7 White                     White                    
 8 Black or African American Black or African American
 9 All other races           Asian                    
10 White                     White                    
# ℹ 19,990 more rows

or select all columns starting with “r”:

yrbss_data %>% 
  select(starts_with("r"))

# A tibble: 20,000 × 3
    record race4                     race7                    
     <dbl> <chr>                     <chr>                    
 1 1317069 Black or African American Black or African American
 2 1317067 White                     White                    
 3 1317066 White                     White                    
 4 1317065 White                     White                    
 5 1317064 White                     White                    
 6 1317063 White                     White                    
 7 1317062 White                     White                    
 8 1317058 Black or African American Black or African American
 9 1317057 All other races           Asian                    
10 1317056 White                     White                    
# ℹ 19,990 more rows

This is a useful shortcut (though use with caution, you need to know your column names and order):

yrbss_data %>% select(1:3)

# A tibble: 20,000 × 3
    record age                   sex   
     <dbl> <chr>                 <chr> 
 1 1317069 18 years old or older Female
 2 1317067 18 years old or older Female
 3 1317066 18 years old or older Female
 4 1317065 18 years old or older Female
 5 1317064 18 years old or older Female
 6 1317063 18 years old or older Female
 7 1317062 18 years old or older Female
 8 1317058 18 years old or older Female
 9 1317057 18 years old or older Female
10 1317056 18 years old or older Female
# ℹ 19,990 more rows

select() and everything() are a useful combination for some quick rearranging of columns:

yrbss_data %>% select(record, bmi, everything())

# A tibble: 20,000 × 8
    record   bmi age                   sex    grade race4         race7 stweight
     <dbl> <dbl> <chr>                 <chr>  <chr> <chr>         <chr>    <dbl>
 1 1317069  24.4 18 years old or older Female 12th  Black or Afr… Blac…     69.0
 2 1317067  18.8 18 years old or older Female 12th  White         White     49.9
 3 1317066  32.1 18 years old or older Female 12th  White         White     90.7
 4 1317065  21.2 18 years old or older Female 12th  White         White     56.2
 5 1317064  23.5 18 years old or older Female 12th  White         White     68.0
 6 1317063  21.7 18 years old or older Female 12th  White         White     52.2
 7 1317062  24.8 18 years old or older Female 12th  White         White     61.2
 8 1317058  17.7 18 years old or older Female 11th  Black or Afr… Blac…     49.9
 9 1317057  19.8 18 years old or older Female 12th  All other ra… Asian     44.4
10 1317056  21.7 18 years old or older Female 12th  White         White     61.2
# ℹ 19,990 more rows

Note there’s also a function relocate() to move around your column order.

19.3 `tidyselect` mini-practice

We can also select variables based on data types
For instance, this use of select(where(is.numeric)) selects all numeric columns
- (where() is a helper function that returns columns where the inside function is TRUE,
- in this case is.numeric returns the TRUE/FALSE)

yrbss_data %>% select(where(is.numeric))

# A tibble: 20,000 × 3
    record   bmi stweight
     <dbl> <dbl>    <dbl>
 1 1317069  24.4     69.0
 2 1317067  18.8     49.9
 3 1317066  32.1     90.7
 4 1317065  21.2     56.2
 5 1317064  23.5     68.0
 6 1317063  21.7     52.2
 7 1317062  24.8     61.2
 8 1317058  17.7     49.9
 9 1317057  19.8     44.4
10 1317056  21.7     61.2
# ℹ 19,990 more rows

Try using is.character to select all character columns, as well as the column bmi.

# change eval: TRUE above
yrbss_data %>% select(____, _____)

19.4 ** Learn more about `tidyselect` (BONUS - on your own)

See some more examples in this slide from a previous BERD workshop
For more info and learning about tidyselect, please run this code in your console:

# install remotes package
install.packages("remotes")

# use remotes to install this package from github
remotes::install_github("laderast/tidyowl")

# load tidyowl package
library(tidyowl)

# interactive tutorial
tidyowl::learn_tidyselect()

19.5 The difference between `filter()` and `select()`

One thing to keep in mind is that:
- filter() works on rows (think FILTER in Excel!), and
- select() works on columns (select your relevant variables)

Keep that in mind!

19.6 Renaming with `rename()`

You can rename columns with select() but it’s a bit easier to do this with the rename() function.
You just need to remember new_name = old_name ordering.
- One equal sign will suffice!

yrbss_data %>% rename(id = record,
                  race_cat = race4,
                  weight = stweight)

# A tibble: 20,000 × 8
        id age                   sex    grade race_cat        race7   bmi weight
     <dbl> <chr>                 <chr>  <chr> <chr>           <chr> <dbl>  <dbl>
 1 1317069 18 years old or older Female 12th  Black or Afric… Blac…  24.4   69.0
 2 1317067 18 years old or older Female 12th  White           White  18.8   49.9
 3 1317066 18 years old or older Female 12th  White           White  32.1   90.7
 4 1317065 18 years old or older Female 12th  White           White  21.2   56.2
 5 1317064 18 years old or older Female 12th  White           White  23.5   68.0
 6 1317063 18 years old or older Female 12th  White           White  21.7   52.2
 7 1317062 18 years old or older Female 12th  White           White  24.8   61.2
 8 1317058 18 years old or older Female 11th  Black or Afric… Blac…  17.7   49.9
 9 1317057 18 years old or older Female 12th  All other races Asian  19.8   44.4
10 1317056 18 years old or older Female 12th  White           White  21.7   61.2
# ℹ 19,990 more rows

Note that we did not save the changes above, and thus we haven’t actually renamed the variables in the dataset yrbss_data!

19.7 Saving our results as a file

Let’s save some processed data in the output/ directory.
We’ll save it as an xlsx file so that we can open it in Excel.

# chain together several data transformation steps with %>%

# use <- to assign the new data to a different name to avoid overwriting the original dataset

processed_data <- yrbss_data %>%   
  select(-race7) %>%
  rename(id = record,
         race_cat = race4,
         weight = stweight) %>%
  filter(sex == "Female")

glimpse(processed_data)

Rows: 9,592
Columns: 7
$ id       <dbl> 1317069, 1317067, 1317066, 1317065, 1317064, 1317063, 1317062…
$ age      <chr> "18 years old or older", "18 years old or older", "18 years o…
$ sex      <chr> "Female", "Female", "Female", "Female", "Female", "Female", "…
$ grade    <chr> "12th", "12th", "12th", "12th", "12th", "12th", "12th", "11th…
$ race_cat <chr> "Black or African American", "White", "White", "White", "Whit…
$ bmi      <dbl> 24.4296, 18.7813, 32.1429, 21.1713, 23.5433, 21.7107, 24.8448…
$ weight   <dbl> 68.95, 49.90, 90.72, 56.25, 68.04, 52.16, 61.24, 49.90, 44.45…

Save the processed_data:

write_xlsx(x = processed_data,
                path = "output/processed_data.xlsx")

20 Practice time! (part 4)

Perform some subsetting functions on the yrbss_data data:

Use filter() to select people who are in 9th or 10th grade and who have BMI less than 25. Save the resulting data as yrbss_filtered tibble.

Use rename() to rename 2 of the column names of yrbss_filtered to something more meaningful.

Use select() to keep only the columns that denote record, sex, grade, bmi, and the 2 renamed columns.

Arrange the data by bmi, starting with the highest value and ending with the least value.

Save the resulting data set in the file data/practice4_data.xslx.

21 Make new variables: `mutate()`

21.1 Using `mutate` to calculate a new variable based on other variables (1/3)

One use for mutate is to do Excel type calculations using other columns in the data.
- For instance, we might want to covert weight to kg, assuming stweight is in lbs
Notice we use = inside mutate, not == or <-.
- We are not making a logical operation, nor are we assigning something to an object name.

yrbss_data %>% 
    mutate(weight_kg = stweight/2.205) %>%
    glimpse()

Rows: 20,000
Columns: 9
$ record    <dbl> 1317069, 1317067, 1317066, 1317065, 1317064, 1317063, 131706…
$ age       <chr> "18 years old or older", "18 years old or older", "18 years …
$ sex       <chr> "Female", "Female", "Female", "Female", "Female", "Female", …
$ grade     <chr> "12th", "12th", "12th", "12th", "12th", "12th", "12th", "11t…
$ race4     <chr> "Black or African American", "White", "White", "White", "Whi…
$ race7     <chr> "Black or African American", "White", "White", "White", "Whi…
$ bmi       <dbl> 24.4296, 18.7813, 32.1429, 21.1713, 23.5433, 21.7107, 24.844…
$ stweight  <dbl> 68.95, 49.90, 90.72, 56.25, 68.04, 52.16, 61.24, 49.90, 44.4…
$ weight_kg <dbl> 31.26984, 22.63039, 41.14286, 25.51020, 30.85714, 23.65533, …

Remember, since we haven’t assigned this new “mutated” data frame to an object name, our work hasn’t been saved.

21.2 Using `mutate` to calculate a new variable based on other variables (2/3)

Below, we save the new data frame/tibble as yrbss_new
- to avoid overwriting our other data set.
- Check out the environment tab for the new tibble.
Notice also how we can mutate multiple variables at one time.
- We can also mutate using multiple columns.
- Here we have created two new weight variables in different scales
- In the next Practice time, you will make a height variable from bmi and weight_kg

yrbss_new <- yrbss_data %>% 
    mutate(weight_kg = stweight/2.205,
           weight_g = weight_kg*1000)

glimpse(yrbss_new)

Rows: 20,000
Columns: 10
$ record    <dbl> 1317069, 1317067, 1317066, 1317065, 1317064, 1317063, 131706…
$ age       <chr> "18 years old or older", "18 years old or older", "18 years …
$ sex       <chr> "Female", "Female", "Female", "Female", "Female", "Female", …
$ grade     <chr> "12th", "12th", "12th", "12th", "12th", "12th", "12th", "11t…
$ race4     <chr> "Black or African American", "White", "White", "White", "Whi…
$ race7     <chr> "Black or African American", "White", "White", "White", "Whi…
$ bmi       <dbl> 24.4296, 18.7813, 32.1429, 21.1713, 23.5433, 21.7107, 24.844…
$ stweight  <dbl> 68.95, 49.90, 90.72, 56.25, 68.04, 52.16, 61.24, 49.90, 44.4…
$ weight_kg <dbl> 31.26984, 22.63039, 41.14286, 25.51020, 30.85714, 23.65533, …
$ weight_g  <dbl> 31269.84, 22630.39, 41142.86, 25510.20, 30857.14, 23655.33, …

21.3 Using `mutate` to calculate a new variable based on other variables (3/3)

Figure examining the newly created variables using mutate:

ggplot(yrbss_new) +
  aes(x = stweight, 
      y = weight_kg) +
  geom_point()

Warning: Removed 6557 rows containing missing values or values outside the scale range
(`geom_point()`).

21.4 Brief aside: categorical data (`factor`s)

One data type that we haven’t yet looked at are factors
- factors are how R represents levels of categorical data.
For the most part, you can use character and factors interchangeably for categorical data.
However, there is one main difference.
- factors define the permissible values in a vector with an argument called levels.
- They also define the order in which these values are displayed.

character_vector <- c("Dog", "Dog", "Cat", "Mouse")   # c() is the concatenate function
class(character_vector)

[1] "character"

tabyl(character_vector)

 character_vector n percent
              Cat 1    0.25
              Dog 2    0.50
            Mouse 1    0.25

The default order of levels is in alphanumeric order.
You can make a character vector into a factor vector by using the factor() function, and supplying an argument called levels.

21.5 Levels of a `factor` variable

The levels of a factor are the permissible values in a factor.
The order of the values in a factor variable determine the order in which the values appear in tables and on the axes in a plot.
You can control the order of the categories in a factor by specifying the order of the categories in the levels argument.
You can find out the levels of the factor with the function levels() or tabyl().

factor_vector <- factor(character_vector)

class(factor_vector)

[1] "factor"

# what are the levels of this vector?
levels(factor_vector)

[1] "Cat"   "Dog"   "Mouse"

tabyl(factor_vector)

 factor_vector n percent
           Cat 1    0.25
           Dog 2    0.50
         Mouse 1    0.25

# specify our own ordering of levels
factor_vector2 <- factor(character_vector, 
                        levels = c("Dog", "Cat", "Mouse"))

levels(factor_vector2)

[1] "Dog"   "Cat"   "Mouse"

tabyl(factor_vector2)

 factor_vector2 n percent
            Dog 2    0.50
            Cat 1    0.25
          Mouse 1    0.25

Being able to specify the ordering is the main reason to use factors, at least in plotting and calculating counts.

21.6 Mini-practice

Change the order of factor_vector to be “Mouse”, “Cat”, “Dog”.

Verify that you did it correctly by calling tabyl() or levels()

factor_vector <- factor(____, 
                        levels = _____)

tabyl(factor_vector)

levels(factor_vector)

21.7 Using `mutate` to transform `character` variables into `factors`

We can also use use mutate() to make a character variable a factor.
Let’s convert age from character into factor:

yrbss_data %>% 
  #create a factor variable
  mutate(age_fac = factor(age))

# A tibble: 20,000 × 9
    record age                   sex    grade race4 race7   bmi stweight age_fac
     <dbl> <chr>                 <chr>  <chr> <chr> <chr> <dbl>    <dbl> <fct>  
 1 1317069 18 years old or older Female 12th  Blac… Blac…  24.4     69.0 18 yea…
 2 1317067 18 years old or older Female 12th  White White  18.8     49.9 18 yea…
 3 1317066 18 years old or older Female 12th  White White  32.1     90.7 18 yea…
 4 1317065 18 years old or older Female 12th  White White  21.2     56.2 18 yea…
 5 1317064 18 years old or older Female 12th  White White  23.5     68.0 18 yea…
 6 1317063 18 years old or older Female 12th  White White  21.7     52.2 18 yea…
 7 1317062 18 years old or older Female 12th  White White  24.8     61.2 18 yea…
 8 1317058 18 years old or older Female 11th  Blac… Blac…  17.7     49.9 18 yea…
 9 1317057 18 years old or older Female 12th  All … Asian  19.8     44.4 18 yea…
10 1317056 18 years old or older Female 12th  White White  21.7     61.2 18 yea…
# ℹ 19,990 more rows

An example with grade:

yrbss_data %>% 
  #create a factor variable
  mutate(grade = factor(grade))

# A tibble: 20,000 × 8
    record age                   sex    grade race4         race7   bmi stweight
     <dbl> <chr>                 <chr>  <fct> <chr>         <chr> <dbl>    <dbl>
 1 1317069 18 years old or older Female 12th  Black or Afr… Blac…  24.4     69.0
 2 1317067 18 years old or older Female 12th  White         White  18.8     49.9
 3 1317066 18 years old or older Female 12th  White         White  32.1     90.7
 4 1317065 18 years old or older Female 12th  White         White  21.2     56.2
 5 1317064 18 years old or older Female 12th  White         White  23.5     68.0
 6 1317063 18 years old or older Female 12th  White         White  21.7     52.2
 7 1317062 18 years old or older Female 12th  White         White  24.8     61.2
 8 1317058 18 years old or older Female 11th  Black or Afr… Blac…  17.7     49.9
 9 1317057 18 years old or older Female 12th  All other ra… Asian  19.8     44.4
10 1317056 18 years old or older Female 12th  White         White  21.7     61.2
# ℹ 19,990 more rows

One thing to notice:
- we are doing something called reassignment in the grade example.
- We’re taking the previous values of our variable (grade),
  - doing something to it (making it a factor), and then reassigning the variable grade to our new set of values.

21.8 `levels` of a `factor` variable revisited

Remember when we wanted to decide the order of the categories of the factor?
- We can assign that order using the levels argument in factor().
- We can check the ordering with tabyl()

yrbss_data %>%
  tabyl(grade)    # Order of grades is not correct

 grade    n percent valid_percent
  10th 4907 0.24535     0.2504338
  11th 4891 0.24455     0.2496172
  12th 4577 0.22885     0.2335919
   9th 5219 0.26095     0.2663570
  <NA>  406 0.02030            NA

Create grade_fac variable to be a factor version of grade with “correct” levels order:

yrbss_data %>% 
  mutate(grade_fac = factor(grade, 
                      levels = c("9th","10th","11th","12th"))) %>%
  tabyl(grade_fac)

 grade_fac    n percent valid_percent
       9th 5219 0.26095     0.2663570
      10th 4907 0.24535     0.2504338
      11th 4891 0.24455     0.2496172
      12th 4577 0.22885     0.2335919
      <NA>  406 0.02030            NA

Notice that the order of the grade categories is now correct

Remember spelling matters!

If we try to use a factor level that doesn’t exist (“10” vs “10th”),
- we won’t have any data in that category, and
- the non-specified levels (“10”) get set to NA missing.

yrbss_data %>% 
  mutate(grade_fac = factor(grade, 
                      levels = c("9th","10","11th","12th"))) %>%
  tabyl(grade_fac)

 grade_fac    n percent valid_percent
       9th 5219 0.26095     0.3553483
        10    0 0.00000     0.0000000
      11th 4891 0.24455     0.3330156
      12th 4577 0.22885     0.3116361
      <NA> 5313 0.26565            NA

22 Practice time! (part 5)

Create a new variable height that calculates height based on bmi and weight_kg. Note the formula for BMI is weight/height^2 (weight is in kilograms and height is in meters). Create a scatterplot of bmi vs height.

Create a new factor variable sex_fac that orders the values of sex as Male, Female.

23 Make categorical variables with `case_when()`

23.1 `case_when()`

Currently the grade variable is a character vector. How do we change these to numeric?
Or, how might we dichotomize a continuous variable such as bmi?
Or, what if we want to combine categories?

First let’s start with a simple binary 2 category variable.

We need to follow the basic pattern for each of our categories inside case_when():
- condition ~ category name
- The left side of the ~ is where we can specify how we define the category based on our column variable names.
- The right side of the ~ is where we can specify the category name (as a character).
- Separate each “case” with a comma.

23.2 `case_when()` example 1

In the example below, we are converting sex to a character variable called female:
- sex == "Female" tests whether the column sex is Female, which is how we define the category 1
  - 1 is the new value, and this is the right side of ~
- sex == "Male" tests whether the column sex is Male, which is how we define the category 0
  - 0 is the new value, and this is the right side of ~
Note how NA is handled here:

yrbss_data %>%
  mutate(female = case_when(
    sex == "Female" ~ 1, # do not forget this comma!
    sex == "Male" ~ 0
  )) %>%
  tabyl(female) # look at the categories

 female     n percent valid_percent
      0 10177 0.50885     0.5147959
      1  9592 0.47960     0.4852041
     NA   231 0.01155            NA

23.3 `case_when()` example 2

In the example below, we are dichotomizing the continuous variable bmi into two categories
We need to do this for each level in our category:

left side defines the condition ~ right side names the category

yrbss_data %>%
  mutate(
    bmi30 = case_when(
      bmi < 30 ~ "<30", # don't forget a comma between cases
      bmi >= 30 ~ "30+"
    )
  ) %>%
  tabyl(bmi30)

 bmi30     n percent valid_percent
   30+  1351 0.06755     0.0997637
   <30 12191 0.60955     0.9002363
  <NA>  6458 0.32290            NA

23.4 `case_when()` example 3

Next we recode bmi to have 3 levels:

<20 (not including exactly 20)
20-30 (not including exactly 30)
30+

Notice the middle category we use an & statement similar to how we used logic in filter

yrbss_data %>%
  mutate(
    bmi3cat = case_when(
      bmi < 20 ~ "<20",
      (bmi >=20) & (bmi < 30) ~ "20-30",
      bmi >= 30 ~ "30+"
    )
  ) %>%
  mutate(bmi3cat = factor(bmi3cat)) %>%   # make it factor after creating categories
  tabyl(bmi3cat)

 bmi3cat    n percent valid_percent
     <20 3167 0.15835     0.2338650
   20-30 9024 0.45120     0.6663713
     30+ 1351 0.06755     0.0997637
    <NA> 6458 0.32290            NA

24 Practice time! (part 6)

Create a new category variable called grade_num in yrbss_data that converts grade to numeric.

Create a new variable called race_cat that relabels race4 to have shorter race category names: W (White), H/L (Hispanic/Latino), B/AA (Black or African American), Other (All other races). Make race_cat a factor variable with levels ordered by the size of the group in descending order.

Create a tabyl of the new variable race_cat and race4 to check your work, including the ordering of the factors.

25 Try what we’ve learned with making figures using `ggplot2`

25.1 Recall boxplots `geom_boxplot()`

Boxplots can be used to compare the distribution of a quantitative variable among categories.
- We need a categorical variable (here, our x axis)
- and a continuous numeric variable, shown here on our y axis.

# creating a boxplot

ggplot(yrbss_data) +
  aes(x = grade, 
      y = bmi) +
  geom_boxplot()

Warning: Removed 6458 rows containing non-finite outside the scale range
(`stat_boxplot()`).

The main differences from the scatterplots we created earlier are the geom type and the variables plotted.
We can change the color similarly to scatterplots.
- However, we map to fill and not color if we want to fill in the box with color:

# adding color
ggplot(yrbss_data) +
  aes(x = grade, 
      y = bmi,
      fill = grade) +
  geom_boxplot()

Warning: Removed 6458 rows containing non-finite outside the scale range
(`stat_boxplot()`).

25.2 Factor levels when plotting

Perhaps we want to change the order of the grade levels in our plot.
This is where factor levels are useful:

yrbss_new <- yrbss_data %>% 
  mutate(grade = factor(grade, levels = c("9th", "10th", "11th", "12th"))) %>%
  drop_na(grade)

ggplot(yrbss_new) +
  aes(x = grade, 
      y = bmi,
      fill = grade) +
  geom_boxplot()

Warning: Removed 6193 rows containing non-finite outside the scale range
(`stat_boxplot()`).

25.3 Faceting our boxplot with `facet_wrap()`

One of the most powerful ways to change a visualization is by faceting.
- We can make multiple plots stratifying by a categorical variable.
To do this, we have to add the facet_wrap() command to our plot.
- We need to specify the variable to facet_wrap - race4 by using the vars() function to specify it as a variable.

# facet_wrap:
ggplot(yrbss_new) +
  aes(x = bmi, 
      y = grade,
      fill = grade) +
  geom_boxplot() +
  facet_wrap(vars(factor(race4)))  #NEW!

Warning: Removed 6193 rows containing non-finite outside the scale range
(`stat_boxplot()`).

Don’t try to facet on a continous numeric variable - it won’t work.
Don’t forget to look at the help documentation (e.g., ?facet_wrap) to learn more about additional ways to customize your plots!

25.4 Faceting our boxplot with `facet_grid()`

Another way to facet our plots is with facet_grid(),
- which lets us select the rows and columns based on categorical variables:

ggplot(yrbss_new) +
  aes(x = bmi, 
      y = sex,
      fill = sex) +
  geom_boxplot() +
  facet_grid(rows = vars(grade),  #NEW!
             cols = vars(race4))

Warning: Removed 6193 rows containing non-finite outside the scale range
(`stat_boxplot()`).

25.5 Mini-practice

Facet the boxplot below by age and grade, don’t forget the vars():

# change eval: TRUE above

ggplot(yrbss_new) +
  
  aes(x = bmi, 
      y = sex,
      fill = sex) +
  
  geom_boxplot() + 
  
  facet_grid(______)

25.6 Back to `mutate`: using `mutate` to replace missing values with `replace_na`

Sometimes we want to fill in missing values with a certain value.
We can use the replace_na() function inside of mutate() to specify this.
For example, suppose if the bmi variable is missing,
- we want to replace it with the average bmi across the entire data set.

yrbss_new <- yrbss_data %>%
  mutate(bmi_filled =
           replace_na(bmi,
                      mean(bmi, na.rm = TRUE)))           
yrbss_new %>% 
  select(contains("bmi")) %>% 
  tail()

# A tibble: 6 × 2
    bmi bmi_filled
  <dbl>      <dbl>
1    NA       23.5
2    NA       23.5
3    NA       23.5
4    NA       23.5
5    NA       23.5
6    NA       23.5

We can also fill in missing values for character vectors
- Here we replace missingness in the grade column with unknown:

yrbss_new <- yrbss_new %>%
    mutate(gradefilled = replace_na(grade, "unknown"))

yrbss_new %>% 
  tabyl(gradefilled, grade)

 gradefilled 10th 11th 12th  9th NA_
        10th 4907    0    0    0   0
        11th    0 4891    0    0   0
        12th    0    0 4577    0   0
         9th    0    0    0 5219   0
     unknown    0    0    0    0 406

26 Practice time! (part 7)

Create a barplot of race4, faceted by sex.

27 Useful keyboard shortcuts

Full list of keyboard shortcuts

action	mac	windows/linux
Run code in qmd (or script)	cmd + enter	ctrl + enter
`<-`	option + -	alt + -
interrupt currently running command	esc	esc
keyboard shortcut help	option + shift + k	alt + shift + k

Practice: Try typing code below in your qmd (with shortcut) and evaluating it (with shortcut):

y <- 5
y